Algorithm for Crack Segmentation of Airport Runway Pavement under Complex Background based on Encoder and Decoder

Abstract

Automatic detection and segmentation of airport pavement cracks has always been the focus of attention of the field management department. Due to the different background, shape, color and size of cracks, traditional methods cannot accurately extract crack information from the road surface image with complex background. Therefore, this paper proposes a deep learning-based image detection method for cracks pixel-level segmentation. The proposed network is an encoder-decoder network structure. The encoder uses VGG19 as the backbone network to extract crack features. A spatial pyramid pooling module is introduced between the encoder and decoder to obtain the global crack information. The hole convolution and multi-loss supervision function are introduced to obtain a larger receptive field and improve the segmentation effect of small cracks. This model can be used for efficient multi-scale feature extraction, aggregation and resolution reconstruction, thereby greatly enhancing the fracture segmentation capability of the network. Compared with traditional image processing and other deep learning-based crack segmentation methods, this algorithm has higher accuracy and generalization ability on complex background airport pavements, making the automatic detection and monitoring of airport pavement more efficient.